When the underlying physical phenomenology (medium, sediment, bottom, etc.) is space-time varying along with corresponding nonstationary statistics characterizing noise and uncertainties, then sequential methods must be applied to capture the underlying processes. Sequential detection and estimation techniques offer distinct advantages over batch methods. A reasonable signal processing approach to solve this class of problem is to employ adaptive or parametrically adaptive signal models and noise to capture these phenomena. In this paper, we develop a sequential approach to solve the signal detection problem in a nonstationary environment.